generate_RNG_streams: Implementation RngStreams for N tasks according to Pierre...
In Burke-Lauenroth-Lab/rSFSW2: Simulation Framework for SOILWAT2

Description Usage Arguments Details Value See Also Examples

The function clusterSetRNGStream creates a stream for each worker/worker, and thus replicability can only be realized if each task is assigned to the same worker on repeated runs. This is usually not guaranteed with load-balancing parallel computations or when a long computation is being re-started from previous partially completed tasks/results. This implementation generates a stream for each unique tasks (instead for each worker) and thus avoids such problems.

1	generate_RNG_streams(N, seed = NULL, reproducible = TRUE)

`N`	An integer. The number of streams to generate.
`seed`	An integer or `NULL`. The seed used by `set.seed` to set the (global/master) random generator, i.e., before generating the seeds of the streams.
`reproducible`	A logical value. If `TRUE`, then `N` are prepared. If `FALSE`, then instead `NA`s are returned.

The current RNG kind, if required, must be captured before this function is called because the function sets the kind to “L'Ecuyer-CMRG” (see examples).

A list of length N containing the seed for each stream.

clusterSetRNGStream

RNGkind_prev <- RNGkind()
seeds <- generate_RNG_streams(10, seed = 123)
# do work with random numbers
RNGkind(kind = RNGkind_prev[1], normal.kind = RNGkind_prev[2])